LLM API pricing by provider
Updated
We track 41 LLM API providers, 1,671 prices in all, each read from the provider's own pricing page (newest check 2026-11-15). Amazon Bedrock lists the most models (108). DeepInfra is the cheapest host for the most shared models (37). 9 providers publish batch prices and 35 publish cached-input prices.
| Provider | Models | Cheapest model (input / output per 1M) | Tiers published | Cheapest for | Checked |
|---|---|---|---|---|---|
| Amazon Bedrock host | 108 | Voxtral Mini 1.0 USD 0.04 / USD 0.04 | cached, batch, flex / spot, priority, fast | 7 of 68 (1 tied) | 2026-11-15 |
| Azure OpenAI and Foundry Models host | 97 | NVIDIA Nemotron 3.5 Lightning 30B A3B USD 0.06 / USD 0.22 | cached, batch, flex / spot, priority, fast | 0 of 53 | 2026-11-14 |
| GMI Cloud host | 97 | DeepSeek V4 Flash USD 0.091 / USD 0.182 | cached | 23 of 91 (5 tied) | 2026-11-15 |
| DeepInfra host | 92 | Llama 3.1 8B Instruct USD 0.02 / USD 0.04 | cached, flex / spot, priority | 37 of 81 (9 tied) | 2026-11-14 |
| Featherless host | 87 | Gemma 3 4B USD 0.05 / USD 0.10 | cached | 13 of 86 (5 tied) | 2026-11-15 |
| Novita AI host | 67 | Llama 3.1 8B Instruct USD 0.02 / USD 0.05 | cached, fast | 6 of 58 | 2026-11-14 |
| Vertex AI host | 63 | GPT OSS 20B USD 0.07 / USD 0.25 | cached, batch, priority | 3 of 54 | 2026-11-14 |
| DigitalOcean Gradient host | 59 | GPT-5 Nano USD 0.05 / USD 0.40 | cached, flex / spot, fast | 0 of 57 | 2026-11-14 |
| Venice host | 57 | Qwen 2.5 7B USD 0.05 / USD 0.13 | cached, fast, ultrafast | 2 of 45 | 2026-11-15 |
| SiliconFlow host | 56 | Qwen 2.5 7B USD 0.05 / USD 0.05 | cached | 11 of 47 | 2026-11-15 |
| Alibaba Cloud Model Studio maker | 42 | Qwen3.7 Flash USD 0.03 / USD 0.13 | cached, fast | 1 of 26 (1 tied) | 2026-11-14 |
| Cloudflare Workers AI host | 40 | Granite 4.0 H Micro USD 0.017 / USD 0.112 | cached | 4 of 29 (1 tied) | 2026-11-14 |
| Parasail host | 39 | GPT OSS 20B USD 0.04 / USD 0.20 | cached, fast | 1 of 33 | 2026-11-14 |
| OpenAI maker | 37 | GPT-5 Nano USD 0.05 / USD 0.40 | cached, batch, flex / spot, fast, ultrafast | 27 of 27 (27 tied) | 2026-11-14 |
| Together AI host | 25 | Rnj-1 Instruct USD 0.15 / USD 0.15 | cached | 1 of 21 | 2026-11-14 |
| Perplexity maker | 24 | Claude Haiku 5.5 USD 0.10 / USD 0.50 | cached | 0 of 22 | 2026-11-15 |
| Z.ai maker | 21 | GLM-4-32B-0414-128K USD 0.10 / USD 0.10 | cached | 4 of 12 (4 tied) | 2026-11-14 |
| CoreWeave Forge (formerly W&B Inference) host | 20 | GPT OSS 20B USD 0.03 / USD 0.13 | cached | 7 of 20 | 2026-11-15 |
| Anthropic maker | 18 | Claude Haiku 5.5 USD 0.10 / USD 0.50 | cached, batch, fast | 15 of 15 (15 tied) | 2026-11-14 |
| Chutes host | 18 | gemma-4-31B-it-turbo USD 0.12 / USD 0.37 | cached, fast | 8 of 18 (1 tied) | 2026-11-15 |
| Fireworks AI host | 15 | NVIDIA Nemotron 3.5 Lightning 30B A3B USD 0.05 / USD 0.20 | cached, flex / spot, priority, fast | 0 of 13 | 2026-11-14 |
| Baidu maker | 14 | DeepSeek V4 Flash USD 0.14 / USD 0.28 | cached, fast | 1 of 11 (1 tied) | 2026-11-15 |
| Baseten host | 14 | DeepSeek V4 Flash 0731 USD 0.13 / USD 0.26 | cached, fast | 2 of 14 (2 tied) | 2026-11-14 |
| Gemini API maker | 12 | Gemini 2.5 Flash-Lite USD 0.10 / USD 0.40 | cached, batch, flex / spot, priority | 11 of 11 (11 tied) | 2026-11-14 |
| Hyperbolic host | 11 | DeepSeek V4 Flash 0731 USD 0.06 / USD 0.18 | cached | 0 of 11 | 2026-11-14 |
| Crusoe Cloud host | 9 | NVIDIA Nemotron 3.5 Lightning 30B A3B USD 0.05 / USD 0.20 | cached | 1 of 8 | 2026-11-15 |
| IBM watsonx.ai maker | 9 | Granite 4.0 H Small USD 0.0636 / USD 0.265 | standard only | 0 of 7 | 2026-11-15 |
| Mistral AI maker | 9 | Ministral 3 3B USD 0.10 / USD 0.10 | cached, batch, priority | 3 of 6 (3 tied) | 2026-11-15 |
| xAI maker | 8 | Grok Build 0.1 USD 1 / USD 2 | cached | 6 of 6 (6 tied) | 2026-11-14 |
| AkashML host | 7 | GPT OSS 20B USD 0.018 / USD 0.09 | cached | 1 of 7 | 2026-11-15 |
| Amazon Nova on Bedrock maker | 7 | Nova Micro 1.0 USD 0.035 / USD 0.14 | batch, flex / spot, priority, fast | — | 2026-11-15 |
| Cohere maker | 6 | Command R7B (12-2024) USD 0.0375 / USD 0.15 | standard only | 1 of 1 | 2026-11-14 |
| FriendliAI host | 5 | Gemma 4 31B USD 0.14 / USD 0.40 | cached | 0 of 5 | 2026-11-15 |
| SambaNova Cloud host | 5 | OpenAI GPT OSS 120B USD 0.22 / USD 0.59 | cached | 0 of 5 | 2026-11-14 |
| Groq host | 4 | GPT OSS 20B USD 0.075 / USD 0.30 | cached | 0 of 4 | 2026-11-14 |
| Moonshot AI (Kimi) maker | 4 | Kimi K2.6 USD 0.95 / USD 4 | cached, batch, fast | 0 of 4 | 2026-11-14 |
| Nscale host | 4 | Qwen3.8-Flash-Next-FP8 USD 0.15 / USD 0.47 | standard only | 0 of 3 | 2026-11-15 |
| AI21 Labs maker | 2 | Jamba Mini 2 USD 0.20 / USD 0.40 | standard only | — | 2026-11-15 |
| Cerebras host | 2 | OpenAI GPT OSS 120B USD 0.35 / USD 0.75 | standard only | 0 of 2 | 2026-11-14 |
| DeepSeek maker | 2 | DeepSeek V4.1 Flash 1017 USD 0.30 / USD 1.20 | cached | 0 of 2 | 2026-11-14 |
| MiniMax maker | 2 | MiniMax M3.1 Flash USD 0.15 / USD 0.60 | cached, priority | 1 of 2 (1 tied) | 2026-11-14 |
| BroadNet Inference this site's owner, not ranked | 4 | See BroadNet pricing | per token, hourly key, flat plans | — | rate card |
"Models": distinct models with a standard per-token price. "Cheapest model": lowest standard price on a blend of 3 input to 1 output token. "Cheapest for": models where the provider has the lowest standard price, out of the models it shares with at least one other provider; "tied" counts models where another provider charges the same (a maker's own API is listed first on a tie). Tiers: cached input, batch, flex or spot, priority and fast, where the provider publishes a per-model price.
Report a wrong price or number (email to info@broadnet.ai, with this page's link filled in).
All models: API prices compared · Cheapest LLM APIs right now · Download every price (CSV, JSON)
Questions
Which LLM API provider is the cheapest?
No single provider is cheapest for every model. Counting models with standard prices from two or more providers, DeepInfra is cheapest for 37, OpenAI is cheapest for 27 and GMI Cloud is cheapest for 23. Each provider page lists the models it wins and by how much, on a blend of 3 input tokens to 1 output token.
Which providers offer batch pricing per model?
Amazon Bedrock, Azure OpenAI and Foundry Models, Vertex AI, OpenAI, Anthropic, Gemini API, Mistral AI, Amazon Nova on Bedrock and Moonshot AI (Kimi) publish a batch price per model. Others, such as Novita AI, DigitalOcean Gradient, Alibaba Cloud Model Studio and Together AI, describe batch only as a blanket discount or not per model, so we list no batch rows for them.
Which providers publish cached-input prices?
35 of 41: Amazon Bedrock, Azure OpenAI and Foundry Models, GMI Cloud, DeepInfra, Featherless, Novita AI, Vertex AI, DigitalOcean Gradient, Venice, SiliconFlow, Alibaba Cloud Model Studio, Cloudflare Workers AI, Parasail, OpenAI, Together AI, Perplexity, Z.ai, CoreWeave Forge (formerly W&B Inference), Anthropic, Chutes, Fireworks AI, Baidu, Baseten, Gemini API, Hyperbolic, Crusoe Cloud, Mistral AI, xAI, AkashML, FriendliAI, SambaNova Cloud, Groq, Moonshot AI (Kimi), DeepSeek and MiniMax. IBM watsonx.ai, Amazon Nova on Bedrock, Cohere, Nscale, AI21 Labs and Cerebras show no cache price on the pages we read, so cached tokens bill at the input price there.
Which providers have priority or flex tiers?
Priority: Amazon Bedrock, Azure OpenAI and Foundry Models, DeepInfra, Vertex AI, Fireworks AI, Gemini API, Mistral AI, Amazon Nova on Bedrock and MiniMax. Flex or spot: Amazon Bedrock, Azure OpenAI and Foundry Models, DeepInfra, DigitalOcean Gradient, OpenAI, Fireworks AI, Gemini API and Amazon Nova on Bedrock. Priority costs more for reliability; flex and spot cost less but can be slow or refused at capacity.
Where do these prices come from?
From each provider's own pricing page or model pages, read with AI assistance and a page renderer, with the URL, check date and a quote stored for every price. No aggregator feed is used. 16 of the 41 providers are model makers selling their own models (some also host others); the rest are hosts.
Read as MarkdownAsk ChatGPTAsk ClaudeAsk Perplexity
Sources
- AI21 Labs: pricing (checked 2026-11-15)
- AkashML: pricing (checked 2026-11-15)
- Alibaba Cloud Model Studio: pricing (checked 2026-11-14)
- Amazon Bedrock: pricing (checked 2026-11-14)
- Amazon Nova (Bedrock): pricing (checked 2026-11-14)
- Anthropic: pricing (checked 2026-11-14)
- Baidu: pricing (checked 2026-11-15)
- Baseten: pricing (checked 2026-11-14)
- Cerebras: pricing (checked 2026-11-14)
- Chutes: pricing (checked 2026-11-15)
- Cloudflare Workers AI: pricing (checked 2026-11-14)
- Cohere: pricing (checked 2026-11-14)
- CoreWeave Forge (formerly W&B Inference): pricing (checked 2026-11-15)
- Crusoe Cloud: pricing (checked 2026-11-15)
- DeepInfra: pricing (checked 2026-11-14)
- DeepSeek: pricing (checked 2026-11-14)
- DigitalOcean Gradient (Serverless Inference): pricing (checked 2026-11-14)
- Featherless: pricing (checked 2026-11-15)
- Fireworks AI: pricing (checked 2026-11-14)
- FriendliAI: pricing (checked 2026-11-15)
- GMI Cloud: pricing (checked 2026-11-15)
- Google (Gemini API): pricing (checked 2026-11-14)
- Google Cloud Vertex AI: pricing (checked 2026-11-14)
- Groq: pricing (checked 2026-11-14)
- Hyperbolic: pricing (checked 2026-11-14)
- IBM watsonx.ai: pricing (checked 2026-11-15)
- Microsoft Azure (Azure OpenAI + Foundry Models): pricing (checked 2026-11-14)
- MiniMax: pricing (checked 2026-11-14)
- Mistral AI: pricing (checked 2026-11-14)
- Moonshot AI (Kimi): pricing (checked 2026-11-14)
- Novita AI: pricing (checked 2026-11-14)
- Nscale: pricing (checked 2026-11-15)
- OpenAI: pricing (checked 2026-11-14)
- Parasail: pricing (checked 2026-11-14)
- Perplexity: pricing (checked 2026-11-15)
- SambaNova Cloud: pricing (checked 2026-11-14)
- SiliconFlow: pricing (checked 2026-11-14)
- Together AI: pricing (checked 2026-11-14)
- Venice: pricing (checked 2026-11-15)
- xAI: pricing (checked 2026-11-14)
- Z.ai: pricing (checked 2026-11-14)